THE 1999 LONG COURSE ON PATHOLOGY OF THE UTERINE CORPUS AND

Advances in Cervical Screening Technology Mark H. Stoler, M.D. Department of Pathology, University of Virginia Health System, Charlottesville, Virginia

take into account that each annual test is an inde- The Pap smear unquestionably is a successful pendent event. Consequently, in an appropriate screening test for . However, recent target population, the risk for missing serious dis- advances in technology have raised questions re- ease with three annual consecutive screenings is garding whether the conventional Pap smear is still approximately 1%. At least half of false-negative the standard of care. This article relates issues of Pap smears are due to inadequate sampling. Either screening and cost-effectiveness to the state of the the cells are not collected or they are not present on art in thin layer preparations, cytology automation, the slide. One third to one half are due to errors in human papillomavirus screening, human papillo- the screening process involving locator or inter- mavirus vaccines, and other cervical screening ad- preter error. Much of the discussion on available juncts. Perhaps nowhere in medicine is clinical de- new technology that follows is focused on lowering cision making being more strongly influenced by the false-negative rate, addressing issues of sam- market and other external forces than in cervical pling, sample preparation, or screening/quality . control. The following background data are derived from KEY WORDS: Advances, Cancer, Cervix, Human several sources, primarily the U.S. Census Bureau, papillomavirus, Neoplasia. the American Cancer Society (ACS), the College of Mod Pathol 2000;13(3):275–284 American Pathologists (CAP), and several online cancer databases (6). They are included to provide some perspective on the advances to be discussed. SCREENING CONCEPTS AND PERSPECTIVES Of the approximately 272 million people in the The Pap smear is the best example of a successful United States, 139 million are female. If we assume cancer screening (1, 2). Since the late 1940s, the that the major target population for cervical cancer coincidence of evolving technology, an accessible screening includes only women between the ages of target organ, and a relatively long preinvasive de- 15 and 70, then the target population size is re- velopment time for this disease with a historically duced to approximately 95 million. A common es- bad outcome provided the impetus for cytologic timate of the volume of Pap smears performed an- screening. As is often noted, the Pap smear has nually is approximately 50 million. Therefore, only never been subjected to a formal clinical trial to approximately half of the population is getting an- determine its efficacy. Rather, early anecdotal pos- nual screening. Of course, this estimate does not itive experience, combined with pressing clinical reflect issues of maldistribution whereby a popula- need, leads to its widespread implementation (3–5). tion subset is highly screened with multiple repeat Approximately 500,000 cases of cervical cancer examinations versus a fairly large subset of poor, occur each year. In contrast to the situation in the uninsured, or ethnic minorities who have never screened populations, cervical cancer still ranks as been screened or are screened inadequately. An the first or second most frequent carcinoma in estimate of the distribution of cytologic diagnoses women (fighting for number 1 with breast cancer) extrapolated to the U.S. population, taking into ac- worldwide. In addition to these features, the effec- count reasonable estimates of the Pap smear false- tiveness of treatment of the preinvasive form of negative fraction per diagnostic category, is pre- cervical cancer combined with the low morbidity of sented in Table 1. treatment facilitated the decrease in cancer in The American Cancer Society estimates that screened populations. The widely quoted 20% there will be 12,800 cases of cervical carcinoma in false-negative grade for a single does not the United States with approximately 4,600 deaths. In unscreened populations, cervical cancer inci- Copyright © 2000 by The United States and Canadian Academy of dence ranges from 30 to 40 in 100,000 to as high as Pathology, Inc. VOL. 13, NO. 3, P. 275, 2000 Printed in the U.S.A. 1 in 1,000, compared with the current U.S. rate of Date of acceptance: November 8, 1999. approximately 5 in 100,000. Hence, screening has

275 TABLE 1. Estimated U.S. Prevalence of Cervical Pathologies

Multiplied by Estimated Corrected for Divided by % Frequency 50 Million %FNF FNF Frequency 0.52 (Females) Pap Smears ASCUS 2.8 20 3.36 1.68Eϩ06 3.23Eϩ06 LSIL 1.97 10 2.17 1.09Eϩ06 2.10Eϩ06 HSIL 0.51 10 0.55 2.75Eϩ05 5.29Eϩ05 SCC 0.026 5 0.027 1.35Eϩ04 2.60Eϩ04 AdCa 0.0046 10 0.005 2.50Eϩ03 4.81Eϩ03 ASCUS:LSIL NA NA NA 1.54 NA ASCUS:SIL NA NA NA 1.23 NA LSIL:HSIL NA NA NA 3.96 NA Data based in part on a CAP study of 1,741,515 smears from 338 laboratories and extrapolated to U.S. population. (Cancer data corrected for American Cancer Society numbers.) FNF, false negative fraction; ASCUS, atypical squamous cells of undetermined significance; L/HSIL, low-/high-grade squamous intraepithelial lesion; SCC, squamous cell carcinoma; AdCa, adenocarcinoma. effected up to a 90% reduction in the incidence of dressed problems of improving sample collection, cervical cancer. Likewise, Gustafsson et al. (7, 8) specimen preparation, screening locator errors, fa- performed a careful study of worldwide databases tigue, and so forth to drive down the false-negative to get an estimate of the incidence rate of cervical fraction while increasing the sensitivity and speci- cancer in various populations before the effects of ficity of this screening test. Some newer technolo- screening. They analyzed more than 83,000 cases, gies attempt to “threaten” the primacy of the Pap and their studies provide a good statistical profile of smear for cervical cancer screening. In the future, cervical cancer worldwide. The mean age at peak given the advances in our understanding of cervical incidence of diagnosis is 54 Ϯ 8 years, and the mean cancer pathogenesis, widespread human papillo- range from the earliest onset of cervical cancer mavirus (HPV) vaccination may make cervical can- (26 Ϯ 3) to peak is 28 Ϯ 7 years. The shape of the cer screening a thing of the past for our children or curve of cancer incidence is interesting and sug- grandchildren. What follows addresses the cur- gests that either exposure to the cancer-causing rently available technologies and the questions of agents decreases with age or target cells are less whether these technologies help us do better in susceptible to malignant transformation with age. cervical cancer screening while trying to balance Compared with the worldwide data, the U.S. and this against a perspective of “if we can do better, is other screened populations have approximately a it worth the cost?” decade earlier mean as well as a questionable rising Recently, Melamed et al. (9) attempted to evalu- incidence, speculated to be due to the effect of ate issues of cost in cervical screening technology. sexual behavior on maturing cohorts. They stated that any cost-effectiveness analysis From Table 1, the pool of all cytologic abnormal- should take a societal as well as a medical perspec- ities (atypical squamous cells of undetermined sig- tive. Assessing the marginal impact of any new nificance [ASCUS] and above) is only 5.9% of the technological advance should be the primary goal population. The corresponding number for low- of such analyses. Certainly, the goal of any new grade squamous intraepithelial lesions (LSIL)ϩϭ technology should be to improve the health of the 2.7% and for high-grade squamous intraepithelial population to which the technology is applied while lesions (HSIL)ϩϭ0.56%. Thus, despite widespread hopefully reducing costs. In reality, there are very “negative” publicity, the Pap smear has done a re- few examples of recently introduced new technol- markable job of finding the “needles in the hay- ogy in which these goals have been achieved. A stack.” This success is despite that the test is still stringent requirement to reduce cost can have a performed very much as originally invented. In- markedly adverse effect on the introduction of any deed, simple but major early advances included the new technology. If price is the only determinant of conversion from a vaginal pool specimen to direct technological implementation, such a strict re- cervical sampling, an improved differential staining quirement may freeze current technology in place. technique, and the training of a highly skilled work- Price plus value, emphasizing the long term, is a force of dedicated cytotechnologists, which have much harder but realistically needed perspective. A combined to bring about the current situation—a full analysis of cost-effectiveness is beyond the Pap test so good that the general population ex- scope of this article (10, 11). However, in a screened pects it to be perfect. population such as in the United States, at least No test is perfect, including the Pap smear. In the 50% of the cancers occur in patients who have past decade, significant technical advances have never had a single screening examination. Another presaged the possibility of improving what is al- 25 to 33% occurs in people who are inadequately ready a very good test. These advances have ad- screened. It should be obvious then, that any im-

276 Modern Pathology provement in screening technology will not signif- ments from some clinical professional organiza- icantly affect the cervical cancer rate because fewer tions that state that new technologies are unproved than 10 to 15% of the cancers occur in patients who or not cost-effective and therefore should not be are being adequately screened even with our cur- implemented. Clearly, the Pap smear is an under- rent “primitive” technology. In addition, the cur- valued test! rent U.S. screening effort combined with the cost of If money cannot be shifted from one available therapy costs approximately $3 to 5 billion a year. A pot to the another, then what alternatives exist for $10 “improvement” equals $1 billion of additional freeing dollars to implement new technologies or to health care cost when applied to the entire popu- reach effectively more of the population? The only lation at risk. Without effectively bringing more pa- obvious answers include screening less frequently tients into the screening pool, a perfect improve- and screening a different proportion of the popu- ment could be expected to save at most 500 to 1,000 lation based on some preselection criteria. Regard- lives at a cost of $1 million or more per case. Is ing the former, Frame and Frame (12) stated that society willing to pay this price? Are there ways of the only significant factors that should determine improving the distribution of the health care dollar the frequency of screening are the “window” of such that new technology may be implemented opportunity reflecting the length of time necessary within the current multibillion-dollar framework? I to develop the target disease and the sensitivity of do not have answers to these questions, but the the test. It is surprising that the incidence of the following observations conclude my remarks on disease in a population does not affect the efficacy screening. Contrast the Pap smear with another of screening. Frame estimated, based on a review of common screening test, serum prostate specific an- the literature, that the overall sensitivity of a single tigen (PSA) analyses. Pap examination is 80%. The window from acqui- Costs for a PSA range between $10 and $15, de- sition of mild cervical dysplasia to the development pending on whether indirect costs are included in of carcinoma in situ and/or invasive cancer is at the analysis. The current charge at many institu- least 6 to 10, if not 20, years. Given that model, tions is approximately $60, although actual reim- screening every 3 years detects 96% of patients who bursement is again variable depending on the are at risk. This model compares well with the in- payer. Even at 50% reimbursement (typical of ternational data used in his study as a comparison Medicare), this test is reimbursed at up to five times group. The analyses clearly show that there is little the rate of a screening Pap smear: This for a test to be gained by screening more frequently than that is nearly completely automated and requires every 3 years. Even recent estimates of only 50% only a blood draw. The benefits of widespread PSA Pap smear sensitivity may not justify annual screening are at best controversial, and the target screening (13). Counterintuitively, more frequent population is much more highly selected. screening may ultimately decrease the sensitivity of In contrast, the current cost for a conventional the given test. This is because frequent screening Pap smear is approximately $10 per test, and the filters out the “easy” cases and leaves cases that are charge for all but the abnormal cases, in which a harder to identify and diagnose in the residual pop- pathologist is involved, is approximately twice that ulation. An examination of their formula and as- at $20 per test. These costs and charges do not sumptions clearly shows that “an ounce of preven- reflect that Pap smears continue to be used as a tion is better than a pound of cure, but 2 ounces is “loss leader” in many institutions. In addition, not better than 1.” Although such an analysis clearly many Pap smears are not billed separately as a points to a potential source of funds for technolog- clinical laboratory test. It is difficult to get exact ical innovation, changing the behavior patterns of numbers, because this is one of the deeply buried populations seems to be a daunting task. secrets in the interaction between pathology labo- Given the above issues, what “advances” are cur- ratories and clinicians. For many patients, the Pap rently available or on the horizon to address the smear is paid for directly by the clinician, usually at problems with Pap smears? The problems, for the a minimal rate. The clinician then charges the pa- most part, are centered on decreasing the false- tient the usual “exam ϩ Pap smear” collection/ negative fraction, thereby increasing sensitivity interpretation fee and pockets the difference. Thus, while hopefully not compromising specificity. there is a strong negative incentive to reimburse Many of the current device-based advances center Pap smears at what even Medicare thinks is a rea- on addressing two issues: optimization of the cel- sonable technical charge, because every cent spent lular sample and improvement in screening the on increased Pap reimbursement is money taken cellular sample for abnormality. Most consumers directly from the clinician’s pocket. In my opinion, do not appreciate the complexity of the Pap smear a significant fraction of the dollars needed to im- screening process. Simple things that can be done plement new technology is tied up in this quagmire, to improve the quality of the cellular sample, in- and such a system calls into question pronounce- cluding appropriate patient preparation (e.g., trying

Advances in Cervical Screening Technology (M.H. Stoler) 277 to avoid sampling at menses), can greatly improve virtually all of the studies are biased against the new the signal-to-noise ratio in the average Pap test. technique in that they are so-called split-sample This article does not analyze the pros and cons of studies in which the conventional Pap smear is different sampling devices. Suffice it to say that the prepared from the cytologic sample first and the amount of material does vary widely with sampling remainder is suspended in the liquid buffer system. device and that sampling devices that adequately More dramatic improvements in sensitivity, partic- collect material across the squamocolumnar junc- ularly for high-grade lesions, as well as specimen tion (transformation zone) improve accuracy. adequacy, have been demonstrated with recent direct-to-vial studies. Some similar preliminary data are available for the AutoCyte Prep. Concerns THIN LAYER TECHNOLOGIES have been raised about the adequacy of the Thin- Prep for the detection of glandular lesions; the orig- The attractions of thin layer technologies include inal FDA trial showed a decrease in specimen ade- improvements in sampling, fixation, staining, and quacy for the presence of an adequate endocervical background. In essence, most of the variables in component. More recent data suggest that these slide quality are removed from the realm of the concerns are an artifact of two processes. Some of sample collector and placed under more stringent the softer cervical “broom” devices may not ade- control within the laboratory. quately sample the transformation zone, particu- The average Pap smear slide contains between larly if inappropriately used. Such devices may also 50,000 and 300,000 cells. Depending on the patient, be inappropriate for certain patients, depending on a significant proportion of these cells may be in- their cervical anatomy. Furthermore, the more re- flammatory cells and/or red cells. In contrast, DNA cent direct-to-vial studies demonstrate significant analysis of the average Pap collection demonstrates improvements in the number of specimens judged that more than 1 million cells are routinely col- to be adequate for an endocervical component. lected in the total sample. Thus, as little as 5% of the The obvious disadvantages of thin layer technol- sample may actually make it onto the slide; with the ogies are the trade-off in sample preparation time traditional Pap sampling method, this is not a ran- and cost in the laboratory. These negatives may be dom subset of the total sample collected. The ad- compensated by an improved screening environ- vantages of thin layer technologies address all of ment with clear cellular visualization, less fatigue, the problems in the preanalytic stage of the Pap and perhaps decreased screening time as a result of smear. The entire sample is placed directly into a the smaller number of cells and smaller cell visual- vial of liquid fixative, optimizing fixation. Random- ization area on the slide. However, the significance ization of the sample ensures a more representative of even a single abnormal cell in these relatively sampling on the slide submitted for screening. Op- decreased cellularity preparations is increased. timization of fixation and staining addresses the Thus, marked decreases in screening time may not significant fraction of ASCUS diagnoses related to be possible as technologists may spend more time obscuring factors or poor fixation. In May 1996, looking at fewer cells to find the small population of Cytyc Corporation received Food and Drug Admin- abnormal cells present on the slide. The other ob- istration (FDA) approval for the ThinPrep 2000 Sys- vious problem is the markedly increased cost asso- tem. As of this writing, ThinPrep is the only mono- ciated with these preparations. Cost-benefit analy- layer system approved for use in gynecologic ses notwithstanding, the FDA has granted Cytyc cytology. A competing technology, AutoCyte Prep claims that allow the ThinPrep to be a replacement (AutoCyte, Inc.) is in the final stages of premarket for the conventional Pap smear and cite that it is approval from the FDA and most likely will be avail- significantly more effective than the conventional able by the time this is published. Pap smear. Currently, the ThinPrep method seems A review of the available literature for the Thin- to have penetrated approximately 10 to 15% of the Prep 2000 System and its predecessor, ThinPrep Pap smear market. Beta, in trials that assessed more than 80,000 sam- ples reveals that the ThinPrep technology produces slides that are superior to a conventional Pap SCREENING AUTOMATION smear’s in finding cases of abnormality (14–28). Improvements in sensitivity are noted primarily for For the past 3 years, two devices, Neopath’s Au- cases of ASCUS and LSIL. However, some studies toPap 300 QC and Neuromedical Systems’ PapNet, have demonstrated improvements in finding cases have been available for the quality control rescreen- of high-grade abnormality as well. No study dem- ing of negative Pap smears. For the past year, the onstrated inferiority for the monolayer technology. AutoPap System has been the only device approved All studies demonstrated at least equivalence, and for primary screening. The PapNet System was un- most demonstrated improvement. Furthermore, dergoing clinical trials as a primary screener (see

278 Modern Pathology below). AutoCyte has a screener that is tightly use in primary screening of Pap smears is some- linked to the AutoCyte Prep method. There are a what different from the system that received ap- few other companies that are still earlier in the proval for quality control rescreening (34–39). Like development process (16, 21, 29). the AutoPap, the new PapNet system places the The AutoPap System allows up to 25% of the screening/scanning instrument in the laboratory. most normal slides that are scanned into the system This solves a major problem with the previous sys- to be sorted out for no further human review. The tem, which required users to ship their slides, often concept that a machine, without human input, can at great distance and expense, to a remote scanning review a cytopathology specimen is potentially a laboratory. The new scanner and software presents major advance in anatomic pathology practice (30– on a computerized review station 128 images for 33). As automated blood smear analyzers had tre- each slide that demonstrate areas with the highest mendous impact on the hematology laboratory, so, likelihood of abnormality. Triage of slides into “no too, could automation have a tremendous impact review” and “review” is determined by the cyto- on the conventional practice of cytopathology. In technologist’s impression on the workstation. It is the FDA clinical trial supporting the use of the important to note that in these trials, thin layer AutoPap, Wilbur et al. (33) conducted a two-armed preparations, as well as conventional Pap smears, prospective study involving 25,000 conventional were also being tested. Likewise, Neopath is con- Pap smears. This study compared AutoPap-assisted ducting clinical trials to remove a major impedi- practice at a 25% sort rate combined with a 15% ment to further implementation of the AutoPap AutoPap-directed quality control rescreen to con- technology, because thin layer preparations and ventional practice including the Clinical Laborato- AutoPap are currently mutually exclusive technol- ries Improvement Act–mandated 10% random re- ogies. However, it should be obvious that the two screening. The ranking scores generated by the technologies greatly complement each other. The AutoPap for the human reviewed fraction were pro- superior cytologic preservation and decrease in vided to the cytotechnologist. This critical part of overlapping cell presentations of thin layer tech- the protocol seems to have modified technologists’ niques can only make the job of these very sophis- behavior/vigilance for the better. The results of this ticated imaging computers much easier. study demonstrate statistically significant improve- The state of screening automation is first gener- ments in the detection of abnormal slides at the ation. Besides the obvious improvements in sort levels of ASCUSϩ, LSIL, and LSILϩ, as well as sta- rates that can potentially be brought about by the tistical equivalence for HSILϩ. In addition, there use of combined thin layer and automation tech- was an average of 16% improvement in the false- nology, other potential improvements are envi- positive rate. Thus, AutoPap-assisted practice not sioned for the near future. These include site- only improved sensitivity but also simultaneously directed screening, whereby the “alarms” by which improved specificity. Neopath’s achievement of these automated systems rank and/or triage slides breaking through the automation barrier with a are communicated to the cytotechnologist in a primary screening instrument in cytopathology is manner that directs him or her to the areas of the highly significant. The potential for maintaining or slide with the highest likelihood of abnormality. increasing capacity in cytopathology laboratories at Such an AutoPap-based system that generates a a time when skilled cytotechnologists are in dimin- “PAPMAP” has already been implemented in Tai- ishing supply yet the government is focusing efforts wan with impressive results. Interfacing such a sys- to increase the number of people availing them- tem with a computer-controlled stage such as Path- selves of the Pap smear screening system should finder (Neopath) could even further improve support its potential use. Of course, new technology efficiency in the screening process. does not come without some increase in cost. How- Of course, all of this technology is contingent on ever, Neopath based the claims granted by the FDA the survival of these companies. As in many new for the use of the AutoPap System as a primary technology arenas, the cytopathology automation screener on a very conservative presentation. The field is rapidly evolving. In the 3 months since the data contained in the primary screening trial may presentation at the USCAP meeting, Neuromedical actually support statistical superiority or equiva- Systems, Inc., has declared bankruptcy. Its intellec- lence at higher sort rates. Overseas implementa- tual property has been sold to Autocyte, Inc. The tions of the AutoPap technology, particularly in ar- field has further consolidated with the announced eas where cytotechnology services are more poorly merger of Autocyte, Inc., and Neopath, Inc. Thus, developed, demonstrate the ability of the system to from four competitors, the marketplace is reduced improve markedly the quality of cytopathology ser- to just two. It would be a shame if the decades of vices while simultaneously decreasing workload. work that have now culminated in actual working The PapNet System that was undergoing clinical cytopathology automation processes were ended trials both in Europe and in the United States for prematurely or terminated inappropriately because

Advances in Cervical Screening Technology (M.H. Stoler) 279 of the severe problems induced by continued un- of 11 to 13 high-risk viruses have been fairly well derreimbursement for Pap smear services. characterized (47–49). Compared with polymerase chain reaction (PCR) analysis using L1 consensus primers, the HCT has a lower sensitivity. However, HPV SCREENING the HCT is more specific for the presence of clini- The estimated overall HPV prevalence in the U.S. cally detectable cervical abnormalities compared target population is approximately 20%. The prev- with PCR, which, because of its higher sensitivity, alence varies greatly with age (Table 2). These data picks up a significantly higher proportion of pa- have implications for a discussion about the utility tients without clinically detectable disease. of HPV testing as a screening procedure. As noted previously, molecular technologies con- It is clear that virtually all lesions encompassed tinue to evolve rapidly. The newest iteration of the by the term “cervical neoplasia” are HPV associ- HCT (HC II, approved in March 1999) is relatively ated. The epidemiologic and molecular evidence semiautomated, uses a microtitre format, and has supporting this is convincing (40, 41). Epidemio- up to 50 times the analytic sensitivity of the current logic studies demonstrate that a positive HPV test is test. Whether the improved sensitivity is of clinical the most powerful independent risk factor for the benefit greatly depends on whether one is using the development of both cervical dysplasia and inva- test for screening or diagnosis/triage and on the sive cancer. Once HPV status is accounted for, the population characteristics. The interplay among relative risk associated with traditional factors, such sensitivity, specificity, and disease prevalence as sexual behavior, becomes insignificant. In lim- needs to be kept in mind when evaluating the utility ited studies, HPV infection precedes and predicts of any test. Likewise, PCR/amplification technolo- for the development of cervical precancer as well as gies are rapidly evolving. In addition, the expanding invasive cancer. Furthermore, virtually 100% of in- sequence database of all relevant HPVs makes it vasive carcinomas from around the world have likely that the new powerful “DNA-chip” technolo- been shown to be associated with a limited spec- gies may replace or augment current HPV testing trum of HPV types (42). methods. Given the strength of these associations, an ob- Might HPV testing be a better screening method? vious question is whether screening for HPV using This question has been most thoroughly examined some sort of molecular diagnostic test would be by workers in the Netherlands, who have proposed superior for selecting the population at risk for can- using an extremely sensitive PCR-based method as cer development (43–46). The answer to this appar- the first step in a cervical cancer screening program ently simple question is, unfortunately, complex. (50–52). If one were designing a cervical cancer Part of the problem is technical. Which HPV test screening program from scratch, this approach should be used? HPV testing, as all molecular diag- makes a tremendous amount of sense. Nearly 100% nostics, is continually evolving. Until recently, there of the pathology of interest is HPV positive. Con- has been only one commercially available FDA- versely, if a sufficiently sensitive screening test approved test for HPV, the Hybrid Capture tube test showed that an individual was not HPV positive, (HCT) marketed by Digene Diagnostics. The sensi- then the incidence of disease would be so low as to tivity, specificity, and predictive values of the make screening nearly worthless. Considering the “tube” test with a 10 pg/ml cutoff value for a group high prevalence of HPVs in the pathology of interest and the relatively long time frame from acquisition of infection until the development of cervical can- TABLE 2. Estimates of Numbers of HPV-Infected Females Based on U.S. Census Data Projections for July cer, the relative value of initial triage based on HPV 1999 and National Cancer Institute Estimates of HPV status is obvious. The lower the prevalence of HPV Positivity by Age in the population to be screened, the better the #of Estimated performance profile of an extremely sensitive HPV Females % of Total Total Age HPV screening test. For instance, the incidence of cervi- Divided by Females HPV/1000 Prevalence 1000 cal cancer in women who are younger than 25 to 30 0–10 19,091 13.71 1.00% 190.91 years is extremely low, and the prevalence of HPV 11–19 19,195 13.78 20.00% 3839 in the United States drops from approximately 40% 20–29 17,864 12.82 40.00% 7145.6 at age 20 to 10 to 20% at age 30 to 40 (or as low as 30–39 21,204 15.22 20.00% 4240.8 40–49 21,020 15.1 10.00% 2102 4 or 5% at age 30 in the Netherlands). Under these 50–59 15,126 10.9 5.00% 756.3 conditions, it may not make sense to spend re- 60–69 10,647 7.65 5.00% 532.35 sources on screening young women, most of whom 70–79 9,188 6.6 5.00% 459.4 80–89 4,796 3.4 5.00% 239.8 develop only transient low-grade lesions. The 90–99 1,105 0.8 5.00% 55.25 Dutch proposal seeks to screen the entire popula- 100ϩ 56 0.04 5.00% 2.8 tion at age 30 with the most sensitive HPV test HPV, human papillomavirus. available combined with a single cytologic screen-

280 Modern Pathology ing. Patients whose results are positive on either vaccine. Vaccines developed in rabbits, cows, and test would be entered into a program of more in- dogs all show great promise. Canine oral - tense routine screening, whereas the “double neg- viruses (COPVs) are effectively prevented by intra- ative” patients would be returned to the general dermal injection in the footpad of either a formalin population pool that would then be screened on a inactivated COPV extract or COPV L1 viruslike long-interval basis of 5 to 10 years. Again, if the particles. Immunization of approximately 60,000 prevalence of detectable virus is low and the dis- beagles during a 3-year period resulted in complete ease prevalence is also low, such a system makes for protection against naturally acquired COPV- extremely rational triage and resource utilization. induced . The arguments become even stronger if the cost Approximately 80% of HPV cancers are associ- and reliability of the HPV test become comparable ated with a limited type spectrum of HPV 16, 18, 31, to cytologic methods. Indeed, in some recent stud- and 45. Several vaccine trials, most initially target- ies, HPV testing seemed more reliable than the Pap ing HPV 16, are undergoing Phase I and Phase II smear because of superior sensitivity in identifying testing. Obviously, the long natural history of both patients with cervical abnormalities. For instance, a HPV infection and cervical cancer, together with large recently published triage study of ASCUS pa- the fact that the optimal target population involves tients evaluated HC II testing for “oncogenic” HPVs young persons before the onset of sexual activity, versus repeat smear as an index for colposcopic complicates the development of such vaccines. referral (53). The sensitivity for HSIL or greater in However, the potential success of an HPV vaccine the HPV testing arm was 89.2% with a specificity of program could produce the first example of true 64.1%. In contrast, the sensitivity for repeat Pap cancer prophylaxis and ultimately lead to the elim- smear was 76.2%. This difference approached sta- ination of the entire cervical cytology screening sys- tistical significance. This and other studies strongly tem. suggest that HPV testing will evolve into routine clinical practice. Furthermore, prospective studies that address a rational basis for HPV primary PAP SMEAR REPLACEMENTS screening are needed and planned in the Nether- lands and at other sites. Whether such a program A growing literature sights the shortcomings of could be tested in the United States is questionable the Pap smear, focusing on the causes of false- because of the relative mobility of the U.S. popula- negative smears and the problem of cervical cancer tion and the strongly ingrained emphasis on annual in the 25 to 33% of individuals who have been Pap smear screening. screened. Much of the previous discussion has been directed at technologies that try to rectify that situation. There are still other approaches (63, 64): HPV VACCINES those that improve direct visualization of the cervix, many in a manner similar to , and those The recognition that HPVs are the primary cause that rely on nonvisual or biophysical analyses that of cervical cancer strongly raises the possibility of seek to distinguish and even grade the normal/ the use of HPV vaccines for both treatment and reactive versus the neoplastic cervix. prophylaxis of cervical cancer (54–62). A successful Among the visualization methods short of full- prophylactic HPV vaccine could virtually eliminate blown colposcopy, the one most widely known is the need for cervical cancer screening programs. (65, 66). This patented technique, This admirable long-term goal is just possibly com- available exclusively through National Testing Lab- ing into reach. oratories (Fenton, MO), is a standardized photo- There are more than 100 types of HPVs, more graph of the cervix under acetic acid–enhanced than 30 of which are prevalent in the cervix. It is visualization, which yields two images that are in- unclear whether immunity to any specific type pro- terpreted centrally by licensed colposcopy experts. vides cross-reactive immunity to other types. Thus, The grading system used has undergone evolution the ultimate vaccine likely may be a complex poly- mainly to improve specificity. Cervicography has valent mixture. Until recently, the lack of an abun- been repeatedly reported to be more sensitive and dant source of HPV antigens markedly impeded less specific than cytology. Note that in my opinion, vaccine development. However, recombinant there is a bias in much of this literature in that the methods that are capable of generating viruslike Pap smear for comparison is taken at the same time particles that contain the HPV L1 and L2 capsid as the cervicogram. A vigorously collected smear proteins have been the major technical advance could affect the quality of the subsequent image. promoting HPV vaccine development. Studies per- Cervicography seems to work best as a Pap smear formed in animals reveal consistent and promising complement, as the two together have a better sen- findings for the development of a prophylactic HPV sitivity/specificity profile than either alone. How-

Advances in Cervical Screening Technology (M.H. Stoler) 281 ever, few data compare the utility of this technique 4. Noller KL. In defense of the Pap smear [editorial]. Obstet with an enhanced Pap smear using automation Gynecol Surv 1998;53(5):261. 5. Patnick J. Has screening for cervical cancer been successful? and/or thin layer methods. The same can be said Br J Obstet Gynaecol 1997;104(8):876–8. for all of the other methods mentioned next. 6. Jones BA. Rescreening in gynecologic cytology: Recreening is a variation of colposcopy that of 3762 previous cases for high grade Squamous Intraepithe- uses blue-white chemiluminescent light along with lial Lesions and Carcinoma—a College of American Pathol- acetic acid and magnification to examine the cervix. ogists Q-Probes study of 312 Institutions. Arch Pathol Lab Very limited published data are available on this Med 1995;119:1097–103. 7. Gustafsson L, Ponten J, Bergstrom R, Adami HO. Interna- technique, which relies on established colposcopic tional incidence rates of invasive cervical cancer before cy- skills (67). Prospective trials on screening popula- tological screening. Int J Cancer 1997;71(2):159–65. tions with appropriate controls have yet to be per- 8. Gustafsson L, Ponten J, Zack M, Adami HO. International formed. incidence rates of invasive cervical cancer after introduction Polarprobe is another adjunctive or potential of cytological screening. Cancer Causes Control 1997;8(5): 755–63. screening technique. The Polarprobe is a portable 9. Melamed MR, Hutchinson ML, Kaufman EA, Schecter CB, optoelectronic device that differentiates tissue Garner D, Kobler TB, et al. Evaluation of costs and benefits types on the basis of voltage decay and multiwave- of advances in cytologic technology: International Academy length light scatter. Readings are taken by applying of Cytology Task Force Summary. Acta Cytol 1998;42(1):69– the probe directly to the surface of the cervix, and 75. 10. Brown AD, Garber AM. Cost-effectiveness of three methods the signals are interpreted immediately by a note- to enhance the sensitivity of Papanicolaou testing. JAMA book computer that relates the patterns to algo- 1999;281(4):347–53. rithms for six tissue types. Limited studies have 11. McMeekin DS, McGonigle KF, Vasilev SA. Cervical cancer shown reasonable concordance with colpohistol- prevention: toward cost-effective screening. Medscape ogy. This device is undergoing trial for FDA ap- Womens Health 1997;2(12):1. 12. Frame PS, Frame JS. Determinants of cancer screening fre- proval as a screening adjunct. The attraction is the quency: the example of screening for cervical cancer. J Am potential for immediate diagnosis, which could Board Fam Pract 1998;11(2):87–95. have significant utility in areas in which cytology 13. McCrory DC, Mather DB, Bastian L, et al. Evaluation of services are lacking. cervical cytology. Evidence report/technology assessment Several other devices are undergoing develop- No. 5. (Prepared by Duke University under Contract No. ment and testing on the basis that, similar to Po- 290-97-0014. AHCPR Publication No. 99-E010.) Rockville, MD: Agency for Health Care Policy and Research; 1999. larprobe, biophysical properties can be measured 14. Austin RM, Ramzy I. Increased detection of epithelial cell in vivo/in situ and can discriminate the healthy abnormalities by liquid-based gynecologic cytology prepa- from the pathologic cervix. Infrared spectroscopy, rations: a review of accumulated data. Acta Cytol 1998;42(1): nuclear magnetic resonance spectroscopy, and 178–84. laser-induced fluorescence all demonstrate some 15. Lee KR, Ashfaq R, Birdsong GG, Corkill ME, McIntosh KM, Inhorn SL. Comparison of conventional Papanicolaou technical basis for clinical trials (68–71). smears and a fluid-based, thin-layer system for cervical can- Obviously, all of these approaches still require cer screening. Obstet Gynecol 1997;90(2):278–84. the patient to have a cervical examination. Clearly, 16. Linder J. Automation in cytopathology. Am J Clin Pathol patients who are willing to submit themselves to 1992;98(4 Suppl 1):S47–S51. cervical screening are not the problem. In under- 17. Linder J, Zahniser D. The ThinPrep Pap test: a review of clinical studies. Acta Cytol 1997;41(1):30–8. developed areas, even simple visual inspection or 18. Linder J. A decade has passed . . . the Pap smear and cervical enhanced inspection with a flashlight after the ap- cancer [editorial; comment]. Am J Clin Pathol 1997;108(5): plication of vinegar can be used to downstage cer- 492–8. vical cancer and improve outcome. The dilemma in 19. Linder J. Recent advances in thin-layer cytology. Diagn Cy- ultimately using any or all of the new technologies topathol 1998;18(1):24–32. is tied to cost-effectiveness analyses and the 20. McNeil C. The new Pap technologies: where they stand. J Natl Cancer Inst 1997;89(6):411. strength of medical arguments for versus the reluc- 21. McNeil C. New Pap test technologies embark on shifting tance of payors to support implementation. seas. J Natl Cancer Inst 1997;89(6):410–12. 22. Papillo JL, Zarka MA, St. John TL. Evaluation of the ThinPrep Pap test in clinical practice: a seven-month, 16,314-case experience in REFERENCES northern Vermont. Acta Cytol 1998;42(1):203–8. 23. Sherman ME, Schiffman MH, Lorincz AT, Herrero R, 1. National Institutes of Health Consensus Development Con- Hutchinson ML, Bratti C, et al. Cervical specimens collected ference statement on cervical cancer, April 1–3, 1996. Gy- in liquid buffer are suitable for both cytologic screening and necol Oncol 1997;66(3):351–61. ancillary human papillomavirus testing. Cancer 1997;81(2): 2. Recommendations for cervical screening 1997. Members of 89–97. the Working Party on Cervical Screening. N Z Med J 1998; 24. Sherman ME, Mendoza M, Lee KR, Ashfaq R, Birdsong GG, 111(1062):94–8. Corkill ME, et al. Performance of liquid-based, thin-layer 3. Herbert A. Is cervical screening working? A cytopathologist’s cervical cytology: correlation with reference diagnoses and view from the United Kingdom [see comments]. Hum Pathol human papillomavirus testing [In Process Citation]. Mod 1997;28(2):120–6. Pathol 1998;11(9):837–43.

282 Modern Pathology 25. Takahashi M, Kimura M, Akagi A, Naitoh M. AutoCyte SCREEN Putaud I, et al. Comparative analysis of human papilloma- interactive automated primary cytology screening system: a pre- virus detection by hybrid capture assay and routine cyto- liminary evaluation. Acta Cytol 1998;42(1):185–8. logic screening to detect high-grade cervical lesions. Int J 26. Wilbur DC, Cibas ES, Merritt S, James LP, Berger BM, Bon- Cancer 1998;75(4):525–8. figlio TA. ThinPrep Processor: clinical trials demonstrate an 45. Jenkins D, Sherlaw-Johnson C, Gallivan S. Can papilloma increased detection rate of abnormal cervical cytologic spec- virus testing be used to improve cervical cancer screening? imens [see comments]. Am J Clin Pathol 1994;101(2):209–14. Int J Cancer 1996;65(6):768–73. 27. Wilbur DC, Dubeshter B, Angel C, Atkison KM. Use of thin- 46. Schneider A. Virologic screening. Eur J Obstet Gynecol Re- layer preparations for gynecologic smears with emphasis on prod Biol 1996;65(1):61–3. the cytomorphology of high-grade intraepithelial lesions and 47. Cope JU, Hildesheim A, Schiffman MH, Manos MM, Lorincz carcinomas. Diagn Cytopathol 1996;14(3):201–11. AT, Burk RD, et al. Comparison of the hybrid capture tube 28. Williamson SL, Hair T, Wadehra V. The effects of different test and PCR for detection of human papillomavirus DNA in sampling techniques on smear quality and the diagnosis of cervical specimens. J Clin Microbiol 1997;35(9):2262–5. cytological abnormalities in cervical screening. Cytopathol- 48. Lorincz AT. Molecular methods for the detection of human ogy 1997;8(3):188–95. papillomavirus infection. Obstet Gynecol Clin North Am 29. Rosenthal DL. Automation and the endangered future of the 1996;23(3):707–30. Pap test. J Natl Cancer Inst 1998;90(10):738–49. 49. Lorincz AT. Hybrid Capture method for detection of human 30. Patten SF Jr, Lee JS, Wilbur DC, Bonfiglio TA, Colgan TJ, papillomavirus DNA in clinical specimens: a tool for clinical Richart RM, et al. The AutoPap 300 QC System multicenter management of equivocal Pap smears and for population clinical trials for use in quality control rescreening of cervical screening. J Obstet Gynaecol Res 1996;22(6):629–36. smears: I. A prospective intended use study. Cancer 1998; 50. Walboomers JM, Husman AM, Snijders PJ, Stel HV, Risse EK, 81(6):337–42. Helmerhorst TJ, et al. Human papillomavirus in false nega- 31. Patten SF Jr, Lee JS, Wilbur DC, Bonfiglio TA, Colgan TJ, tive archival cervical smears: implications for screening for Richart RM, et al. The AutoPap 300 QC System multicenter cervical cancer. J Clin Pathol 1995;48(8):728–32. clinical trials for use in quality control rescreening of cervical 51. van Ballegooijen M, van den Akker-van Marle ME, Warmer- smears: II. Prospective and archival sensitivity studies. Can- dam PG, Meijer CJ, Walboomers JM, Habbema JD. Present cer 1997;81(6):343–7. evidence on the value of HPV testing for cervical cancer screen- 32. Wilbur DC, Bonfiglio TA, Rutkowski MA, Atkison KM, Richart ing: a model-based exploration of the (cost-)effectiveness. Br J RM, Lee JS, et al. Sensitivity of the AutoPap 300 QC System Cancer 1997;76(5):651–7. for cervical cytologic abnormalities: biopsy data confirma- 52. Jacobs MV, Snijders PJ, van den Brule AJ, Helmerhorst TJ, tion. Acta Cytol 1996;40(1):127–32. Meijer CJ, Walboomers JM. A general primer GP5ϩ/GP6(ϩ)- 33. Wilbur DC, Prey MU, Miller WM, Pawlick GF, Colgan TJ. The mediated PCR-enzyme immunoassay method for rapid de- AutoPap system for primary screening in cervical cytology: tection of 14 high-risk and 6 low-risk human papillomavirus comparing the results of a prospective, intended-use study genotypes in cervical scrapings. J Clin Microbiol 1997;35(3): with routine manual practice. Acta Cytol 1998;42(1):214–20. 791–5. 34. Mango LJ, Radensky PW. Interactive neural-network- 53. Manos MM, Kinney WK, Hurley LB, Sherman ME, Shieh- assisted screening: a clinical assessment. Acta Cytol 1998; Ngai J, Kurman RJ, et al. Identifying women with cervical 42(1):233–45. neoplasia: using human papillomavirus DNA testing for 35. Duggan MA, Brasher P. Paired comparison of manual and equivocal Papanicolaou results [see comments]. JAMA 1999; automated Pap test screening using the PAPNET system. 281(17):1605–10. Diagn Cytopathol 1997;17(4):248–54. 54. Frazer IH. The role of vaccines in the control of STDs: HPV 36. Radensky PW, Mango LJ. Interactive neural-network- vaccines. Genitourin Med 1996;72(6):398–403. assisted screening: an economic assessment. Acta Cytol 55. Gross G. Therapy of human papillomavirus infection and 1998;42(1):246–52. associated epithelial tumors. Intervirology 1997;40(5–6):368–77. 37. Michelow PM, Hlongwane NF, Leiman G. Simulation of pri- 56. Hines JF, Ghim SJ, Jenson AB. Prospects for human papillo- mary cervical cancer screening by the PAPNET system in an mavirus vaccine development: emerging HPV vaccines. Curr unscreened, high-risk community. Acta Cytol 1997;41(1):88– Opin Obstet Gynecol 1998;10(1):15–9. 92. 57. Lowy DR, Schiller JT. Papillomaviruses and cervical cancer: 38. Team PPM. Assessment of automated primary screening on pathogenesis and vaccine development. J Natl Cancer Inst PAPNET of cervical smears in the prismatic trial. Lancet Monogr 1998;23:27–30. 1999;353:1381–5. 58. McNeil C. HPV vaccines for cervical cancer move toward clinic, 39. Kreuger FAF, van Ballegooijen M, Doornewaard H. Is PAP- encounter social issues. J Natl Cancer Inst 1997;89(22):1664–6. NET suitable for primary screening? Lancet 1999;353:1374–5. 59. McNeil C. HPV vaccine treatment trials proliferate, diversify. 40. Stoler MH. The biology of papillomaviruses. Pathol Case Rev J Natl Cancer Inst 1997;89(4):280–1. 1997;2(1):1–13. 60. Munoz N, Bosch FX. The causal link between HPV and 41. Richart RM, Masood S, Syrjanen KJ, Vassilakos P, Kaufman cervical cancer and its implications for prevention of cervical RH, Meisels A, et al. Human papillomavirus: International cancer. Bull Pan Am Health Org 1996;30(4):362–77. Academy of Cytology Task Force summary. Diagnostic Cy- 61. Sherman ME, Schiffman MH, Strickler H, Hildesheim A. tology Towards the 21st Century: An International Expert Prospects for a prophylactic HPV vaccine: rationale and fu- Conference and Tutorial. Acta Cytol 1998;42(1):50–8. ture implications for cervical cancer screening. Diagn Cyto- 42. Bosch FX, Manos MM, Munoz N, Sherman M, Jansen AM, pathol 1998;18(1):5–9. Peto J, et al. Prevalence of human papillomavirus in cervical 62. Steller MA, Schiller JT. Human papillomavirus immunology cancer: a worldwide perspective. International biological and vaccine prospects. J Natl Cancer Inst Monogr 1996;21:145–8. study on cervical cancer (IBSCC) Study Group [see com- 63. Schneider A, Zahm DM. New adjunctive methods for cervi- ments]. J Nat Cancer Inst 1995;87(11):796–802. cal cancer screening. Obstet Gynecol Clin North Am 1996; 43. Coutlee F, Mayrand MH, Provencher D, Franco E. The future 23(3):657–73. of HPV testing in clinical laboratories and applied virology 64. Spitzer M. Cervical screening adjuncts: recent advances. research. Clin Diagn Virol 1997;8(2):123–41. Am J Obstet Gynecol 1998;179(2):544–56. 44. Clavel C, Bory JP, Rihet S, Masure M, Duval-Binninger I, 65. Baldauf JJ, Dreyfus M, Ritter J, Meyer P, Philippe E. Cervi-

Advances in Cervical Screening Technology (M.H. Stoler) 283 cography. Does it improve cervical cancer screening? Acta potential for non-invasive, automated diagnosis of cervical Cytol 1997;41(2):295–301. intraepithelial neoplasia and use as a surrogate endpoint 66. van Niekerk WA, Dunton CJ, Richart RM, Hilgarth M, Kato H, biomarker. J Cell Biochem Suppl 1994;19:111–9. Kaufman RH, et al. Colposcopy, cervicography, speculos- 69. Ramanujam N, Mitchell MF, Mahadevan A, Thomsen S, Silva copy and endoscopy: International Academy of Cytology E, Richards-Kortum R. Fluorescence spectroscopy: a diag- Task Force summary. Diagnostic Cytology Towards the 21st nostic tool for cervical intraepithelial neoplasia (CIN). Gy- Century: An International Expert Conference and Tutorial. necol Oncol 1994;52(1):31–8. Acta Cytol 1998;42(1):33–49. 70. Mahadevan-Jansen A, Mitchell MF, Ramanujam N, Malpica 67. Loiudice L, Abbiati R, Boselli F, Cecchini G, Costa S, Grossi E, A, Thomsen S, Utzinger U, et al. Near-infrared Raman spec- et al. Improvement of Pap smear sensitivity using a visual troscopy for in vitro detection of cervical precancers. Pho- adjunctive procedure: a co-operative Italian study on specu- tochem Photobiol 1998;68(1):123–32. loscopy (GISPE) [In Process Citation]. Eur J Cancer Prev 71. Lee JH, Cho KS, Kim YM, Kim ST, Mun CW, Na JH, et al. 1998;7(4):295–304. Localized in vivo 1H nuclear MR spectroscopy for evaluation 68. Richards-Kortum R, Mitchell MF, Ramanujam N, Ma- of human uterine cervical carcinoma. AJR Am J Roentgenol hadevan A, Thomsen S. In vivo fluorescence spectroscopy: 1998;170(5):1279–82.

Book Review

Brody JS: The Lung: Molecular Basis of Dis- ter 1 reviews basic concepts of DNA and South- ease, 218 pp, Philadelphia, W.B. Saunders ern, Northern, and Western blots. The Company, 1998 ($52). subsequent chapters on tuberculosis, alveolar ␣ proteinosis, 1-antitrypsin, alveolar proteinosis, In recent years, molecular biologic techniques deficiency, cystic fibrosis, lung cancer, and AIDS have entered into the diagnostic clinical arena are used to explain and illustrate basic molecular and are now powerful tools in the evaluation of concepts and analysis such as polymerase chain neoplasms, infections, and genetic disorders, to reaction, DNA cloning, linkage analysis, onco- name a few. Because of the rapid evolution and genes, and so forth. complexity of molecular biology, this area may Each chapter is clearly and concisely written seem intimidating to practicing physicians and students. and contains multiple tables and figures to make As the author states in the preface, the goal complex material very understandable. The bib- of this book is “to provide sufficient background liography for each chapter is relatively short but to enable physicians and students to understand seems to contain a sufficient number of perti- how molecular biology is now impacting on pul- nent recent references. This text is well suited for monary practice and to be able to understand students and practicing physicians (particularly the coming advances.” This concisely written, pulmonologists) who desire a basic introduction well-illustrated book largely achieves this goal to molecular biology. and is an understandable introduction to molec- ular biology. Anita Dixon The book is organized into 10 chapters fol- Veterans Affairs Medical Center lowed by a glossary of basic terminology. Chap- Kansas City, Missouri

284 Modern Pathology